Computing services are increasingly supplied by remotely-located providers who host various software-based services on networks of computers on behalf of clients. Although convenient for the client of such services, it may be challenging to detect or diagnose issues that may occur during the provision of the service. Various approaches to detecting or diagnosing issues may rely on the collection and dissemination of diagnostic data. For example, operational logs, performance metrics, and trace data may be collected during the provision of a computing service and subsequently reviewed. However, this data may be expensive to generate, transmit, and store.